Data · Global · 10 min read
By RWA Radar Research · Published
Key Takeaways
Open almost any deck pitching tokenization, stablecoins or blockchain payments and you will meet a familiar sentence: cross-border payments cost 4-6%, they are slow and opaque, and the new rail will fix it. The figure is usually waved at “the IMF” or simply left unattributed, as if it were common knowledge.
The number is not made up. But as it travels it quietly mutates, and three things go wrong at once. The cost of one narrow product — retail remittances — gets relabelled as the cost of all cross-border payments. The original source — the World Bank— gets swapped for “the IMF”. And the technology that the source actually models gets swapped for whatever the deck is selling. This page traces the figure back to primary documents and pins down what it really measures — consistent with our primary-source methodology.
The canonical measurement of remittance cost is the World Bank’s Remittance Prices Worldwide (RPW), a quarterly database that tracks the total cost of sending the equivalent of $200 across hundreds of country corridors. In its most recent issue (Issue 53, March 2025, covering Q1 2025), RPW reports the headline number plainly:
In Q1 2025, the Global Average cost for sending remittances was 6.49 percent.
That is the “~6%” you keep seeing — measured as sender fees plus the foreign-exchange margin on a $200 transfer, averaged across corridors (per the World Bank, RPW Issue 53). The figure had risen from 6.26% the prior quarter, so it is not even falling smoothly. And the cost depends heavily on how you send: per the same report, digital remittances averaged 4.85% while non-digital averaged 7.16%, and banks remained the most expensive channel at 14.55% (per the World Bank).
So where does “the IMF” come in? The IMF does use the number — but it cites the World Bank for it. In its February 2025 Fintech Note Estimating the Impact of Digital Money on Cross-Border Flows, the IMF writes that, relying on “the highly detailed database from the World Bank,” remittance costs across roughly 100 mostly low-income countries average 6.2 percent of the amount sent (per the IMF, Fintech Note 2025/002). The IMF is a secondary citationof a World Bank figure. Attributing “4-6%” to “the IMF” is, at best, citing the messenger.
Here is the conflation that matters most. “Remittances” and “cross-border payments” are not synonyms — remittances are a tiny, unusually expensive slice. The same IMF note breaks the market into segments by 2023 flows and average total cost (per the IMF, Fintech Note 2025/002, Table 1):
Read that table once and the marketing slogan collapses. The IMF summarises wholesale costs, citing McKinsey (2018), at about 0.1%, and the broader retail range, drawing on the G20 Roadmap data, at roughly 1.5% to 6%(per the IMF, Fintech Note 2025/002). The “4-6%” sits at the topof that range — it is the remittance corner, not the centre of gravity. Saying “cross-border payments cost 4-6%” is like quoting the price of the most expensive seat and calling it the average ticket.
The IMF is explicit about the consequence: even a large drop in transaction costs would do little to global cross-border volumesin the short run “as a result of the low transaction costs of the wholesale segment” — the savings, the note says, “are relatively more important for remittances” (per the IMF, Fintech Note 2025/002). The friction the slogan points at is real, but it lives in one specific place.
The other figure that gets garbled is the goal. The “3%” you often hear is the remittance target from UN Sustainable Development Goal 10.c, whose text is precise: “By 2030, reduce to less than 3 per cent the transaction costs of migrant remittances and eliminate remittance corridors with costs higher than 5 per cent” (per the UN). The same 3%-by-2030 remittance target is reaffirmed in the G20 Targets for Enhancing Cross-Border Payments.
But the G20 sets a different cost target for retailpayments — “global average cost of payment to be no more than 1%, with no corridors with costs higher than 3% by end-2027” — and for wholesale payments it sets no cost target at all(per the FSB). So the policy world is tracking three separate cost lines; decks that mash them into a single “cut payments from 6% to 3%” story are merging targets that apply to different products on different deadlines.
None of this means the technology is irrelevant. The same IMF note runs an illustrative scenario assuming a 60% reduction in transaction costs from digital-money innovation, and finds the gains land squarely on remittances: a 60% cut on high-cost remittance flows could save roughly $17 billion, about 3.7% of those flows (per the IMF, Fintech Note 2025/002). That is a real, meaningful number for migrant workers — and it is exactly where the “6%” cost lives.
But notice what the IMF is — and is not — modelling. The 60% figure is attributed to digital money broadly(CBDCs, digital payment innovations and, longer-term, tokenization of assets on programmable platforms), not to “DeFi” specifically. The note never claims permissionless DeFi as the comparison; the mechanism it cites is more competition and new intermediaries (per the IMF, Fintech Note 2025/002). When a deck attributes “the IMF says DeFi cuts the 6% cost,” it has substituted a different technology for the one in the source.
How to cite this number honestly, then:
That is the discipline behind every figure on this site, and behind our wider work — see, for instance, our Hong Kong RWA regulation timeline. A number is only as good as the question it answers; this one answers “what does it cost a migrant worker to send $200 home?” — and nothing larger.
World Bank
Financial Stability Board
United Nations